Method and system for managing distributed data

Details Method and system for managing distributed data Method and system for managing distributed data

1998-05-15

2001-01-30

Harrell, Robert B. (Department: 2758)

Electrical computers and digital processing systems: multicomput

Distributed data processing

C709S202000, C709S203000, C709S217000, C709S219000, C709S223000, C707S793000

Reexamination Certificate

active

06182111

ABSTRACT:

BACKGROUND OF THE INVENTION
The present invention relates to a computer system, and more particularly to a method and system for managing distributed data, suitable particularly for the world wide web (WWW), in which a plurality of computers interconnected by a network distribute, share and exchange data in an information system.
First, several terms used in the following description will be explained.
An information system such as WWW and anonymous FTP on the Internet is configured as a “client-server system” which is one type of distributed computer systems. In the client-server system, the processes in the whole system are classified into two parts. The first part is executed by a program (hereinafter called a process) called a “server”, and the second part is executed by processes called “clients”. A client in the information system generally runs on a computer operated by a home user or a company user. The server in the information system stores information to be supplied to clients. The client in the information system stores new information in the server or requests information from the server.
It is common in a computer system that the same information is temporarily copied to a plurality of sites in order to access the information at high speed or increase a possibility of accessibility. Such a copy is discriminably called hint, cache, replica, stash and the like (refer to a document “Distributed Systems (1st ed.) compiled by Sape Mullender, pp. 13-15, ACM press, 1989). In the following, these copies are collectively called a “cache”. To make a cache is called “cache”.
A WWW server in WWW stores information to be serviced, in a unit called a “home page”. Each home page has a name called URL (abbreviation for uniform resource locator). URL is a character string capable of designating a protocol used in WWW, a host name of a computer as an information source, and specific data in the information source. For example, “http://www.hitachi.co.jp/index.html” is a URL.
Generally, each URL is in correspondence with a collection of data including character and image data of a home page. In the following, a collection of such data is called “URL corresponding information” or “URL contents”. A second URL contained in the first URL corresponding information is called a “hyper text link (or simply link)”. That the first URL corresponding information contains a second URL is hereinafter described as “there is a link from the first URL to the second URL”.
The techniques (hereinafter called prior example 1) used by WWW will be explained in the following.
A user of a WWW client supplies a URL of the home page to be accessed, to the WWW server. In the first process type between a WWW server and client, the WWW client requests the WWW server designated by the second element of URL to transmit the home page of URL. In response to this request, the WWW server supplies the home page to the WWW client.
In the second process type, instead of requesting the WWW server designated by the second element of URL supplied from the user, the WWW client requests a second server called a “proxy server” to transmit the home page. The second server acquires the home page of URL from the first WWW server or requests another proxy server to acquire the URL corresponding information. At the repetitive stage of proxy server requests, these proxy servers have parent-child relationships. Proxy servers having a parent-child relationship are described, for example, in a document “A Hierarchical Internet Object Cache” by A. Chankhunthod, et.al., 1996 USENIX Technical Conference, pp. 153-163, 1996.
A WWW client and proxy server can have caches. A cache of a client stores home pages the client acquired in the past, and can be used only by this client. A cache of a proxy server stores home pages acquired by the proxy server in response to a request from one or more clients, a request from another or more other servers, or a request from both, and can be shared by the clients using this proxy server or by this proxy server itself.
The Network News System (hereinafter called prior example 2) is described, for example, in a document “Network News Transfer Protocol: A proposed Standard for the Stream-Based Transmission of News” by B. Kantor, et.al., Network Working Group RFC-977. This system is configured by one or more servers. Generally, a user selects one of the servers by using its client. The information unit in the Network News System is called “news”. Generally, a user supplies news to the server by using its client, and acquires news from the server. As the user supplies news to a first server, the first server sends a copy of the news to a second server, and the second server supplies a copy of the news to another server, and so on. Finally, the copy of the news is supplied to all the servers.
Next, the global area name service, Domain Name System (hereinafter called prior example 3, abbreviated as DNS) will be explained. DNS is described, for example, in a document “Domain Names-Implementation and Specification” by P. Mockapetris, Network Working Group RFC-1035, particularly in the second section thereof. DNS has a correspondence mainly between a symbolic host name and host related information (IP address and mail address). A plurality of DNS servers have a tree structure. A request from a client is processed by tracing the tree structure and transferring the request to a plurality of servers. A resolver which is a DNS client requests the host related information corresponding to a host name to one of the servers. This server returns the host related information corresponding to the host name back to the client, or transfers the request to a parent server of this server (a DNS server nearer to the root of the DSN server tree structure from this server). The parent server grasps which of its child servers have what host related information. Therefore, after the request is transferred to the root of the tree structure, the request is then transferred downward the tree structure to the DNS server capable of processing the request. The request finally reaches the DNS server having the host related information from which the host related information is returned to the client, or alternatively a failure is returned back to the client if every DNS server cannot supply the host related information corresponding to the host name while the request is transferred downward the tree structure.
A method (hereinafter called prior example 4) is also known in which a space of caches is shared by a plurality of computers in a distributed file system of a local area network (LAN). According to a document “Cooperative Caching: Using Remote Client Memory to Improve File System Performance” by Michael Dahlin, et.al., First USENIX Symposium on Operating Systems Design and Implementation, pp. 267-280, 1994, a client first requests for a file block to a server called a “manager”. The manager grasps which file block is stored in what computer. The manager informs the client of the computer which stores the file block, or transfers the request of the client to the computer. Similar methods are known as described in a document “Implementing Global Memory Management in an Workstation Cluster” by M. J. Feeley, et.al., ACM 15th Symposium on Operating Systems Principles, pp. 201-212, 1995, or in a document “Efficient Cooperative Caching Using Hints” by P. Sarkar, et.al., Second USENIX Symposium on Operating Systems Design and Implementation, pp. 35-46, 1996. A plurality of managers can be provided. However, a correspondence between file blocks and managers is prefixed and is known by all clients and servers. This correspondence does not change during system running.
Techniques used by computers called cache-coherent non-uniform memory access (CC-NUMA) and cache only memory access (COMA) will be explained by using following prior examples 5 and 6. The CC-NUMA computer or COMA computer has a mechanism of maintaining coherency between memory fragments (cache lines) cached near at a number of processors. The following two methods are known in p

Affiliated with

Also associated with

Rating

Method and system for managing distributed data does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Method and system for managing distributed data, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for managing distributed data will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2529213